The function of the serotonin transporter (SERT) is to terminate neurotransmission by reuptake of serotonin (5HT) from the synapse into the pre-synaptic nerve terminal. Dysfunction of SERT has been implicated in several neurological diseases including depression, anxiety, mood disorders, and attention deficit hyperactivity disorders (ADHD; Freissmuth et al., 2017). Moreover, drugs of abuse like cocaine or amphetamine interfere with normal SERT function and lead to depletion of the 5HT pools in the pre-synaptic nerve terminal (Hilber et al., 2005).

The structure of SERT has been resolved in the outward-open, outward-occluded, and inward-open conformation (Coleman et al., 2016, 2019). The sodium binding sites were first identified in the homologous bacterial small amino acid transporter LeuT (Yamashita et al., 2005), which also revealed the conserved fold of this transporter family. The substrate binding site (labeled S1) is located in the center of the transporter, halfway through the membrane. The transport cycle leading to 5HT uptake is initiated by binding of substrate and co-transported ions to the outward-open conformation. Full assembly of the transport complex consisting of bound ions and 5HT leads first to 5HT occlusion in the substrate binding site S1, followed by a transition to the inward-open conformation from which substrate and ions are released into the cytosol. Return to the outward-open state is facilitated by binding of a potassium ion or a proton (Nelson and Rudnick, 1979; Hasenhuetl et al., 2016).

An energy source is necessary for facilitating uphill transport, to guaranty directionality and for enabling efficient neurotransmitter clearance by reuptake (Grouleff et al., 2015). SERT belongs to the SLC6 protein family which uses the transmembrane electrochemical gradient of sodium as a primary energy source (Chen et al., 2004). Strict coupling is required between binding of substrate and ions and the key conformational changes of the transport cycle, thereby allowing for efficient transport and initiating transport only once the transport complex has assembled (Tavoulari et al., 2016). Several studies showed that sodium binding stabilizes the outward-open conformation, thereby preventing futile cycling events (Claxton et al., 2010; Zhao and Noskov, 2011; Zhao et al., 2011; Stolzenberg et al., 2015; Tavoulari et al., 2016; Coleman et al., 2019; Li et al., 2019). The high external sodium concentration of 150 mM ensures that the two sodium binding sites are filled, as the K_M of sodium for serotonin transport is 25 mM (Quick, 2003). Moreover, a positive cooperativity between sodium and substrate binding was observed that strongly increases the affinity of 5HT to SERT (Chen and Reith, 2003; Hasenhuetl et al., 2018).

The sodium binding sites (referred to as NA1 for site 1 and NA2 for site 2, see Figure 1) of SERT are sodium selective (Felts et al., 2014), a property shared with LeuT (Yamashita et al., 2005; Noskov and Roux, 2008; Zhao and Noskov, 2011). Accordingly, transport is not sustained by other monovalent ions, while mutations of the ion coordinating residues reduce transport activity (Tavoulari et al., 2009; Andersen et al., 2010; Felts et al., 2014). Structures of SLC6 transporters suggest that NA1 and NA2 have distinct roles, with NA1 mainly responsible for substrate binding, while the ion in NA2 is important for stabilization of the outward-open state (Grouleff et al., 2015; Coleman et al., 2019). Solvation of sodium in NA2 is suggested to be important for the transition to the inward-open state (Forrest et al., 2008; Zhao and Noskov, 2011; Borre et al., 2014; Razavi et al., 2017). Simulations investigating the effect of bound ions started with bound ions or investigated the ion-free state (Zhao and Noskov, 2011; Grouleff et al., 2015; Zomot et al., 2015; Razavi et al., 2017; Li et al., 2019).

Kinetic models of the transport cycle, largely build on electrophysiological measurements, predict a sodium association rate constant of 10⁶ to 10⁷ M⁻¹s⁻¹ (Hasenhuetl et al., 2016, 2018; Burtscher et al., 2019) to the chloride bound outward-open SERT, therefore indicating that at physiological conditions the sodium binding is fast enough (nanosecond to microsecond time-range) to be amendable for direct investigation using molecular dynamics simulation. Starting from a set of 51 SERT structure in the outward-open conformations, extracted from a 0.5 μs long simulation of a sodium free SERT, we simulated sodium binding in 51 independent 150 ns long trajectories (Figure 1), resulting in 8 μs of total simulation time. Sodium binds to at least one of the sodium binding sites in the majority of simulations. We find a transient binding site at the extracellular gate (R104–E493) that acts as initial engagement site with the vestibule of SERT. Sodium enters fast into the vestibule, attracted by a strong negative electrostatic field created by SERT that protrudes into the extracellular medium to facilitate sodium attraction. Before reaching the NA1 and NA2 sites, sodium binds to a second temporary interaction site in the S1 just outside NA1 and NA2. Analyses of the geometries of NA1 and NA2 reveal that their local structures are dynamic in the absence of sodium. The change in geometry of NA1 and NA2 upon sodium binding is reminiscent of an induced-fit mechanism that leads to a more compact, well-defined, and rigid conformation.

We selected the outward-open human SERT crystal structure (PDB ID: 5I71; Coleman et al., 2016) to investigate sodium binding. Missing side chains and the originally absent Cl⁻ ion were positioned using MODELLER 9.20 (Shen and Sali, 2006; Webb and Sali, 2014) creating 100 structures. The best model based on the DOPE score was used for simulations.

The all atom model was converted into a coarse grain representation of the MARTINI force field (Monticelli et al., 2008; de Jong et al., 2013; Wassenaar et al., 2015) and inserted in a 1-palmitoyl-2-oleoylphosphatidylcholine: cholesterol containing membrane (POPC:CHOL 70:30 mol%; van Meer, 1998). The system was solvated in water and 150 mM NaCl. The coarse-grain system was simulated for 1 μs while restraining the protein structure to allow the membrane to accommodate the transporter and to equilibrate around SERT. After membrane equilibration, the coarse-grained system was converted to an all-atom representation (Wassenaar et al., 2014) and the original SERT model replaced the converted protein model to avoid spurious local structural problems induced by the double coordinate conversion of the protocol. Possible atom overlaps between reinserted transporter and the relaxed environment were relaxed using the membed procedure (Wolf et al., 2010). We used the amber ff99SB-ILDN force field (Lindorff-Larsen et al., 2010) to describe SERT, ions and the solvent, and Slipid (Jämbeck and Lyubartsev, 2012, 2013) for POPC and cholesterol. Residue Glu508 was protonated as suggested by structural analysis of the SERT crystal structure (Coleman et al., 2016). All simulations were carried out with GROMACS version 2019.2 (Abraham et al., 2015). The final assembled system was energy-minimized and equilibrated in four steps of 2.5 ns, each by stepwise releasing the position restraints (1,000, 100, 10, 1 kJ/mol/nm) that are active on the Cα atoms and the bound Cl⁻. The production run was carried for 500 ns after removing all position restraints. The temperature was maintained at 310 K using the v-rescale (τ = 0.5 ps) thermostat (Bussi et al., 2007), while separately coupling protein, membrane, and solvent. Pressure was maintained at 1 bar using the Parrinello-Rahman barostat (Parrinello and Rahman, 1981) in a semiisotropic manner and applying a coupling constant of 20.1 ps. Long range electrostatic interactions were described using the smooth particle mesh Ewald method (Darden et al., 1993) applying a cutoff of 0.9 nm. The van der Waals interactions were described using the Lennard Jones potentials applying a cutoff of 0.9 nm. Long-range corrections for energy and pressure were applied. Coordinates of all atoms were recorded every 5 ps. The complete set of parameters of the production run can be found in the Supplementary Materials. The resulting trajectory was used as parent trajectory to extract starting structures for the sodium binding simulations, namely taking a snapshot every 10 ns including time = 0 ns, providing 51 starting structures, after removing the sodium ion that entered the substrate binding pocket S1. The resulting 51 systems were simulated for 150 ns with the same simulation parameters as the parent trajectories.

Figures and statistical analyses were generated by the GROMACS package, R, and python scripts using the MD Analysis package, v0.19.2 (Michaud-Agrawal et al., 2011; Gowers et al., 2016). For visualization VMD (Humphrey et al., 1996) v1.9.3 and Pymol v1.8.4 were used.

The binding and co-transport of extracellular sodium into the cell and the movement downhill its electrochemical gradient is the primary determinant for driving the transport cycle of SERT (Chen et al., 2004). This electrochemical gradient of sodium leads under physiological conditions to the transport of 5HT into cells even during cellular conditions where the transport is uphill the 5HT chemical gradient (Grouleff et al., 2015). The binding of sodium ions stabilizes the outward-open conformation of SERT (Claxton et al., 2010; Zhao and Noskov, 2011; Zhao et al., 2011; Stolzenberg et al., 2015; Tavoulari et al., 2016; Coleman et al., 2019; Li et al., 2019) and strongly increases the affinity of the substrate 5HT, while transport occurs only once the complex between SERT, 5HT and the co-transported ions is established (Masson et al., 1999; Felts et al., 2014; Tavoulari et al., 2016). Stabilization of substrate in the S1 was linked to NA1, as in LeuT the substrate interacts directly with the sodium ion in NA1 (Yamashita et al., 2005; Grouleff et al., 2015; Coleman et al., 2019), while in the monoamine transporters, the sodium ion in NA1 positions the sidechain of the adjacent aspartate (D98 in SERT) to interact with the amino groups of the substrate (Wang et al., 2015; Coleman et al., 2016, 2019). The main role of the sodium ion bound to NA2 is to stabilize the closed intracellular gate and thus the outward-open conformation (Zhao and Noskov, 2011; Zhao et al., 2012; Khelashvili et al., 2015; Tavoulari et al., 2016; Razavi et al., 2017; LeVine et al., 2018).

In this study, we investigate the first step of the transport cycle, which is the binding of sodium ions to the outward-open conformation of SERT in the presence of chloride ions, as experimental data indicate that chloride remains continuously bound throughout the transport cycle (Buchmayer et al., 2013; Hasenhuetl et al., 2016). We find that the structure of SERT is optimized for an efficient recruitment of sodium ions into the open vestibule and observe an association rate constant (k_on) of 9.8·10⁶ M⁻¹s⁻¹, which is in good agreement with experimentally observed rate constants (Hasenhuetl et al., 2016, 2018; Burtscher et al., 2019). The fast binding is induced by the strong electrostatic field of SERT, which has a negative sign in the outer vestibule and guides the positively charged sodium ions into the S1. The computationally identified rate constant should be considered an upper estimate, because in simulations the starting structure of SERT is already in a sodium binding competent conformation, while the experimental procedure of the electrophysiological measurements provides a more complex readout, which is convoluted by conformational changes of SERT. The situation is less clear for LeuT, as sodium, which has a low mM affinity for LeuT (Zhao et al., 2010), was stably bound to NA2 in most simulations (Zhao et al., 2012; Tavoulari et al., 2016), but was also found to dissociate from LeuT through the outer vestibule in one study (Zomot et al., 2015).

The binding of sodium ions to SERT is a multilayered process: sodium ions become initially attracted to a binding zone outside the extracellular salt bridge that serves as an initial recruitment zone for positively charged ions thereby increasing their local concentration. The most important residue in this recruitment zone is D328. A mutation of this residue was shown to decrease the affinity of amphetamines and to reduce amphetamine-mediated efflux (Kortagere et al., 2013). Mutation of the corresponding residue in the dopamine transporter (D313) was shown to decrease the affinity for dopamine, which was suggested to be an indirect effect that is caused by a reduced accessibility of sodium ions for reaching the sodium binding sites (Chen and Reith, 2003).

At the extracellular gate, we find a transient sodium binding site, which shows a high propensity to attract sodium ions that were initially recruited to the zone next to D328. Despite simulations showing high local ion density, sodium ions did not remain bound for very long to this site next to the glutamate residues 493 and 494. A weak sodium binding affinity seems mechanistic important because preventing any interference with proper assembly of the transport complex consisting of ion- and 5HT-bound SERT. If in contrast, this site would represent a strong sodium binding site halfway to the S1, it would be detrimental for transport, because preventing or at least slowing the processes of sodium and 5HT binding. Importantly, the narrow shape of the outer vestibule at the extracellular gate provides sodium ions with only two escape routes: unbinding towards the extracellular space or proceeding towards S1. The direction of the electrostatic field lines pointing towards the more negative S1 ensures that sodium ions move more likely towards NA1 and NA2 if the sites are not yet sodium bound.

In the last step before stably binding to SERT, sodium ions associate with a second transient binding site in the S1 juxtaposed to the NA1 and the NA2 sites. A comparable behavior was observed in LeuT (Zomot et al., 2015). This second transient binding site coincides with the position of positively charged nitrogen atoms of substrates and inhibitors as observed in structures of dDAT and hSERT (Wang et al., 2015; Coleman et al., 2016), suggesting that it has a dual role: (i) to serve as a transient site to efficiently recruit sodium ions; and (ii) to form an interaction site for the positive changed nitrogen of the monoamine substrate 5HT.

Structural changes induced by ligand binding have frequently been described as following an induced-fit mechanism (Koshland et al., 1966) or conformation selection (Monod et al., 1965), which represent extreme views of structural adaptations of proteins to the presence of ligands. The conformational changes associated with sodium binding to NA1 and NA2 suggest an overall induced-fit mechanism (Figure 4), as NA1 and NA2 are more compact in the presence of sodium. The situation is particularly clear for NA1, which is very dynamic and wide open in the absence of sodium, but also NA2 shows a comparable behavior. The time point of sodium association correlates with a very compact geometry of NA1 in the absence of sodium. Such a selective binding process would be consistent with a conformational selection as sodium ions seem to select a particular conformation for their association. At the same time, the NA1 compacts further once a sodium ion is bound. This becomes apparent in Figure 4 as NA1 becomes more compact after sodium binding, thus a clear indication of an induced-fit effect. Together these data indicate that the structural arrangement of NA1 might be a combination of an initial conformational selection followed by an induced-fit condensation. The driving forces for compacting the NA1 site are most likely electrostatic attractions between sodium and the coordinating oxygen atoms of SERT that carry partial negative charges. Water molecules occupy NA1 and NA2 if their site are not binding a sodium ion. The neutral water molecule, which is slightly bigger than a sodium ion, cannot lead to the same electrostatic attractions, while also unable to form hydrogen bonds with all potential interaction partners. For the initial conformational selection, similar arguments may apply: (i) initial stabilization of sodium is more efficient, if the site is more compact, (ii) while the presence of water molecules might be energetically less favorable if the site is more compact.

Beyond the core three carbonyl backbone oxygen atoms of residues G94, V97, and L434 of NA2, the side chains of D437 and S438 contribute to complete the NA2 site. Their interaction with the bound sodium ion is more dynamic and can be replaced by a water-bridged interaction. The conformation of S438 is associated with sodium entry, because the side chain needs to rotate for allowing initial sodium association, thereby forming a small barrier which could explain the slower association rate as compared to NA1. The side chain of residue D437 is partially solvated in our simulations as also observed in other studies (Zhao and Noskov, 2011; Khelashvili et al., 2015; Tavoulari et al., 2016; Razavi et al., 2017; LeVine et al., 2018). These water molecules reach residue D437 from the intracellular side; it is, therefore, conceivable that water interactions with and the dynamics of the side chain of residue D437 are associated with the transition towards the inward-facing state. Both residues (D437 and S438) are not essential for protein function as their mutation does not affect the surface expression of SERT, while transport activity is reduced between 15% and 93%, depending on the identity of the mutation (Felts et al., 2014). These data therefore suggest that the three backbone carbonyl oxygen atoms are essential for sodium binding, while the side chains of D437 and S438 have a secondary role in NA2 function.

In conclusion, our large-scale simulation approach to directly study the association of sodium ions reveal that SERT creates a strong electrostatic field that reaches into the extracellular milieu and attracts positively charged ions towards their binding sites. The data show that specific sites for sodium interactions exist within the outer vestibule which are transient enough to support efficient recruitment and transition of sodium ions. The binding of the sodium ions to the NA1 and the NA2 sites is associated with local structural changes that consist of an initial conformation selection followed by an electrostatic driven induced-fit process.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

DS and TS contributed in the conception of the work, the design of data analysis, the interpretation of the results, and drafted the manuscript. DS performed the simulations, data collection, and developed the analysis code. TS provided funding and supervised the work. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.